Multi-hazard Bridge Design Criteria
June 12, 2017
AASHTO SCOBS MEETING
George C. LeeJerry Shen
Tom Murphy
2
Outline
• Challenges: From LRFD to Multi-hazard LRFD
• Multi-hazard Design Guide Development
• Roadmap for Future Study
3
Bridge Design Specifications
WSD
LFDLRFD (non-extreme loads)
MH-LRFD (all loads)
MH-LRFD (performance-based)
Sustainability Design
Present
Past, Present, and Moving Forward
4
The Goals for the Multi-hazard Bridge Design Development
• What? – Risk-based design
– Straightforward implementation
– Consistent with current LRFD
• Why? – Need more consistent safety
– Need less waste in over-design
– Need ease of future development/implementation
5
FHWA Studies on Consistent Multi-hazard Design
• Study Carried out by MCEER– Pilot Study, 2007-2008
• Monograph stating the present problems and potential approaches
– Framework Development and Survey, 2008-2014
• Principles for considering all hazards on a consistent basis
• Procedures for Multi-hazard calibration
• Current Study Led by Genex– Applies the Framework for calibration of select Limit States
07’ 08’ 09’ 10’ 11’ 12’ 13’ 14’ 15’ 16’
Pilot Study Framework development
Applicationof the Framework
17’
6
Current Project: Multi-hazard Bridge Design Criteria
• Period: 9/2014 - 6/2017
• Research Team– Genex Systems (Contact: Jerry Shen)
– MCEER
– Modjeski and Masters
– Arora and Associates
– FHWA: Starting: W. Philip YenCurrent: Sheila Duwadi (COR)
7
Relevant Projects• FHWA Multi-hazard Design Research 2011
• NCHRP 12-48 (Extreme Hazard)
• NCHRP 12-49 (Seismic)
• NCHRP 12-33 (LRFD Calibration)
• NCHRP 24-31, 24-35 (Foundation Calibration)
• NCHRP 12-36, 12-47, 12-86 (Redundancy)
• NCHRP 24-34 (Scour)
• FHWA Identification of Redundancy Factor Modifiers (ongoing)
8
Extreme Events
• Working Definition: Those limit states where we allow the structure to exhibit behavior beyond that expected at the strength and service limit states.
• Have not generally been calibrated to achieve any specific reliability.
• Limited current guidance on which loads/conditions to combine.
9
Limit State—defined deterministically
10
Prob
abilit
y D
ensi
ty F
unct
ion
Q
Probability of failure
Q = Load
R
R = Resistance
R-Q
R-Q
To increase bridge reliability, i.e., reduce probability of failure1. Increase Load Factor 2. Reduce Resistance Factor
Qn
Rn
AASHTO LRFD Design Limit State Equations
: Load factor: Resistance factor
11
Prob
abilit
y D
ensi
ty F
unct
ion
Q
Probability of failure
Q = Load
R
R = Resistance
R-Q
R-Q
Qn
Rn
To increase bridge reliability, i.e., reduce probability of failure1. Increase Load Factor 2. Reduce Resistance Factor
Qn
Rn
AASHTO LRFD Design Limit State Equations
: Load factor: Resistance factor
12
Prob
abilit
y D
ensi
ty F
unct
ion
Q
Probability of failure
Q = Load
R
R = Resistance
R-Q
R-Q
Qn
To increase bridge reliability, i.e., reduce probability of failure1. Increase Load Factor 2. Reduce Resistance Factor
Rn
Rn
AASHTO LRFD Design Limit State Equations
: Load factor: Resistance factor
13
LRFD with Extreme Events (Multi-hazard)• Reliability of LRFD is calibrated for dead load and live load
applied on superstructure.
• Reliability for other loads and components are NOT fully calibrated.
• Extreme Events: higher risk, but less frequent.
Load Combination Limit States
Dead Load*
Live Load
Water Load
Wind Load on Bridge
Wind Load on
Truck
Use One of These at a TimeEarth-quake Blast Ice
LoadVehicularCollision
VesselCollision
Strength I 1.25 1.75 1.00 -- -- -- -- -- -- --Strength II 1.25 1.35 1.00 -- -- -- -- -- -- --Strength III 1.25 -- 1.00 1.40 -- -- -- -- -- --Strength IV 1.50 -- 1.00 -- -- -- -- -- -- --Strength V 1.25 1.35 1.00 0.40 1.00 -- -- -- -- --Service I 1.00 1.00 1.00 0.30 1.00 -- -- -- -- --Service II 1.00 1.30 1.00 -- -- -- -- -- -- --Service III 1.00 0.80 1.00 -- -- -- -- -- -- --Service IV 1.00 -- 1.00 0.70 -- -- -- -- -- --
Extreme Event I 1.25 0/0.5 1.00 -- -- 1.00 -- -- -- --Extreme Event II 1.25 0.5 1.00 -- -- -- 1.00 1.00 1.00 1.00
Calibrated for superstructure
Not calibrated &Unknown risk
14
Challenges in the Probabilistic Analysis
• Characteristics of loads– Time-variable vs. time-invariable
– Correlation among loads/conditions and resistance
• Redundancy (System Reliability)
• Variation in practice
• Multidisciplinary design considerations (structure, geotech, hydraulics, seismic, etc.)
• Simplicity of design formulas
15
Correlation of Time-Variable Loads
Q2
Q1
∪ 1 2
Q2
Q1
∪ 1 2
Correlated Loads Independent Loads
16
Variations in Practice
• Example: foundation design with scour consideration
Long term degradation+ Contraction scour
Local scour(complex pier)
Soil
Bed after scour
Bed after scour
Local scour(single pier)
Design option 1 Design option 2
Required pile length
(a) (b) (c)
17
Procedure for Code Writer
Equation for Bridge Designer
Track 2B
Track 2A
Track 1
Nominal Loads, Resistance and Scour (Qn, Rn & ys )
Deterministic Analysis
Reliability () of Current Bridges
Design equations
Nominal Loads, Resistance and Scour (Qn, Rn & ys )
Target Calibrated Load and Resistance Factors
Statistical Models
Q
Probability of failure
Q
R
R
R-Q
R-Q
Qn Rn
Qs
Qs
Statistical Models
Q
Probability of failure
Q
R
R
R-Q
R-Q
Qn Rn
Qs
Qs
Calibration Approach
′ Σ ′ ′
Calibrated Reliability
18
Calibration Cases• Calibration1—Strength I
– DL + LL ∪ SC
– Current Design: Full factored dead load and live load with 1.0 total scour from design flood
– 90 Bridges of three bridge spans, three hydrology uncertainties, two types of piles and five design methods
• Calibration 2—Extreme Event I
– DL + LL ∪ SC ∪ EQ
– Current Design: Full factored dead load ~0.5 live load with 1.0 or less total scour from design flood (scour varies among states) and 1.0 earthquake
– Vertical Forces on Piles
– Lateral Forces on Piles
Demo Calibration Case –DL + LL SC EQ
Pile Loads and Resistance
20
• Seismic load from plastic hinge—capacity protection design
Seismic Loading
Gravity and moment from EQ
Horizontal force
21
Variation of Design Cases• Pile capacity method
– Meyerhof (Sand)
– Nordlund (Sand)
– -Tomlinson (Clay)
– method (Clay)
– method (Clay)
• Displacement pile and non-displacement pile
• Bridge span (varies DL/LL and scour components)
• Hydrological uncertainty (varies scour)
DL+LL+SC+EQ
22
• Identify possible failure mechanisms– Different failure mechanisms can lead to different limit
state equations.• Force demand
• Capacity protection
• Displacement demand
Design Limit State Equations—Vertical
ARSSR SR n D D L L E SQ SQ E(1 ) (1 )q R Q Q q Q
C C C D D D
colΔ SΔ SΔ f Y p EΔ SΔ SΔ D[(1 ) ] (1 )q q
ARSn n D D L L E SQ SQ E(1 )R S Q Q q Q
S n n D D L L E E(1 )S R Q Q Q
Vertical
Lateral
Vertical
Lateral
Lateral
)( EELLDDn QQQSRn 1.225
1 0.0222 1
23
• Option 1: Constant Reliability for all Methods (Target 2.00)No Constraint on Factors
• Option 2: Constant Reliability for all Methods (Target 2.00)Fix 1.00, 1.25, 0.50
• Option 3: Constant Reliability for all Methods (Target 2.00)Fix 1.00, 1.25, 0.50, 0.00
• Option 4: Constant Reliability for all Methods (Target 2.00)Fix 1.00, 1.25, 0.50, 1.00
• Option 5: Constant Reliability for all Methods (Target 2.00)Fix 1.00, 1.25, 0.50, 1.00
• Option 6: Constant Reliability for all Methods (Target 2.00)Fix 1.00, 1.25, 0.50, 1.50
• Option 7: Different Reliability for each Methods (Target Varies)Fix 1.00, 1.25, 0.50, 1.50
Recommended Live Load and Scour Factors—Vertical
Recommended
Method Nordlund(Sand)
α-Tomlinson(Clay)
H-Pile 0.835 0.773Concrete-Pile 0.802 0.761
24
• Reliability for proposed bridge design equation
, , ,
Reduced Variation of Reliability—Vertical
Method Nordlund(Sand)
α-Tomlinson(Clay)
H-Pile0.835 0.773
Target 0.95 2.80
Concrete-Pile
0.802 0.761
Target 0.95 2.80
Total Length of Concrete Piles
in ClayCurrent AASHTO
DLSEProposed
DLSE
Case 1 38.6 34.6
Case 4 59.9 63.4
Case 7 73.5 79.5
1 2 3 4 5 6 7 8 90
1
2
2.8
3.5
4
Case #
Rel
iabl
ity In
dex
Re-calculated Reliability Index (H-Pile, -Tomlinson-Method, Clay)
Recalculated Original
Target : 2.80Recalculated :Mean: 2.80COV: 1.93%
1 2 3 4 5 6 7 8 90
1
2
2.8
3.5
4
Case #
Rel
iabl
ity In
dex
Re-calculated Reliability Index (H-Pile, -Tomlinson-Method, Clay)
Recalculated Original
Target : 2.80Recalculated :Mean: 2.80COV: 1.93%
25
• Potential DLSE formulation for lateral force
Seismic Load on a Pile—Lateral (moment)
Potential hinge
Potential hinge
Potential hinge
Potential hinge
26
• Potential DLSE formulation for lateral force
Seismic Load on a Pile—Lateral (shear)
27
• Track 2A - Calibrate the reliability for current bridge design - Lateral
Seismic Load on a Pile—Lateral
28
Findings/Accomplishments in this Project• Probabilistic analysis tools based on MCEER
Framework were developed in this study to consider the complex cases in multi-hazard LRFD calibration.
• Strength I and Extreme Event I Limit States were calibrated for deep (pile) foundation with consideration of scour.
• Scour exhibited significant conservativeness in the vertical load and bending moment on piles, but not in shear load on piles.
• More calibration cases needs to be done considering all practical design details in multiple disciplines.
29
Roadmap• A maximum set of calibration cases was composed.
• Unnecessary calibration cases were eliminated when possible.
• Keep formulas simple as possible.
• May need to repeat when new data/technology is available
• Prioritization
30
Roadmap for Future StudyDimensions of the Task Matrix
Bridge Components Loads• Bridge Elements
− Beam-Slab− Truss− Piers, Abutments and Walls− Foundations
• System/Sub system
• Load Components− Moment− Shear− Axial Force− Displacement
• Failure Mechanism / Performance Level− Rupture of longitudinal rebars− Concrete crushing− Buckling− Soil failure (displacement limit)− Unseating− Shear bar failure
• Load Combinations− DL LL ∪ EQ ∪ SC− DL SC− DL LL ∪ EQ− DL LL ∪ CV ∪ SC− …
31
Data Guidelines• Guidance for loads and resistance
• Important properties:– Model uncertainties and randomness from nature
– Time-dependent /Time-independent intensity
– Correlation
– Annual rate of events
– Duration of an event
• Sample Immediate Needs:– Resistance models
– Scour distribution
32
Recommended Future Effort• The complete calibration of LRFD will benefit the
bridge owners by offering better basis for decision making.
• It is recommended to set up a plan to gradually complete the calibration of LRFD.
• Bridge design specification is a living document that continues taking advantage of new data and technology.– A small cross-disciplinary team working with the
SCOBS and bridge community continuously
– 5 year may be a good calibration/planning cycle
33
Acknowledgement• Oversight committee
– Bruce Johnson, Oregon DOT (Co-chair)
– Susan Hida, California DOT
– Richard Pratt, Alaska DOT
– Wahid Albert, New York DOT
– Bijan Khaleghi, Washington DOT
– Wassem Dekelbab, NCHRP
– Phil Yen, FHWA (Former Chair)
– Sheila Duwadi, FHWA (Chair)
• Meeting/Documentation
– Richard Land, GPI
– Eric Thorkildsen, GPI